CN111222665B - Cloud manufacturing service combination optimization selection method based on preference NSGA-III algorithm - Google Patents
Cloud manufacturing service combination optimization selection method based on preference NSGA-III algorithm Download PDFInfo
- Publication number
- CN111222665B CN111222665B CN201811416182.5A CN201811416182A CN111222665B CN 111222665 B CN111222665 B CN 111222665B CN 201811416182 A CN201811416182 A CN 201811416182A CN 111222665 B CN111222665 B CN 111222665B
- Authority
- CN
- China
- Prior art keywords
- individuals
- service
- algorithm
- individual
- preference
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000004519 manufacturing process Methods 0.000 title claims abstract description 46
- 238000005457 optimization Methods 0.000 title claims abstract description 34
- 238000010187 selection method Methods 0.000 title claims abstract description 21
- 238000000034 method Methods 0.000 claims abstract description 43
- 230000008569 process Effects 0.000 claims abstract description 10
- 238000010276 construction Methods 0.000 claims abstract description 6
- 238000011156 evaluation Methods 0.000 claims description 18
- 238000004220 aggregation Methods 0.000 claims description 10
- 230000002776 aggregation Effects 0.000 claims description 10
- 239000000203 mixture Substances 0.000 claims description 7
- 238000010845 search algorithm Methods 0.000 claims description 7
- 230000014759 maintenance of location Effects 0.000 claims description 6
- 238000013210 evaluation model Methods 0.000 claims description 4
- 230000035772 mutation Effects 0.000 claims description 3
- 230000004931 aggregating effect Effects 0.000 claims description 2
- 239000000463 material Substances 0.000 claims description 2
- 230000002068 genetic effect Effects 0.000 abstract description 5
- 230000002452 interceptive effect Effects 0.000 abstract 1
- 238000012545 processing Methods 0.000 description 7
- 230000006870 function Effects 0.000 description 4
- 238000011160 research Methods 0.000 description 4
- 238000009826 distribution Methods 0.000 description 3
- 238000012384 transportation and delivery Methods 0.000 description 3
- 230000002159 abnormal effect Effects 0.000 description 2
- 238000010586 diagram Methods 0.000 description 2
- 239000004973 liquid crystal related substance Substances 0.000 description 2
- 238000007726 management method Methods 0.000 description 2
- 238000013507 mapping Methods 0.000 description 2
- 102100025912 Melanopsin Human genes 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013278 delphi method Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000004880 explosion Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 239000013213 metal-organic polyhedra Substances 0.000 description 1
- 238000012011 method of payment Methods 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 239000002245 particle Substances 0.000 description 1
- 238000005192 partition Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 238000012797 qualification Methods 0.000 description 1
- 238000013139 quantization Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 238000012552 review Methods 0.000 description 1
- 238000004088 simulation Methods 0.000 description 1
- 238000010561 standard procedure Methods 0.000 description 1
- 238000005303 weighing Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Information and communication technology [ICT] specially adapted for implementation of business processes of specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Biophysics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- General Physics & Mathematics (AREA)
- Economics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Bioinformatics & Computational Biology (AREA)
- Marketing (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Tourism & Hospitality (AREA)
- General Health & Medical Sciences (AREA)
- General Business, Economics & Management (AREA)
- Evolutionary Biology (AREA)
- Operations Research (AREA)
- Artificial Intelligence (AREA)
- Manufacturing & Machinery (AREA)
- Quality & Reliability (AREA)
- Entrepreneurship & Innovation (AREA)
- Game Theory and Decision Science (AREA)
- Development Economics (AREA)
- Physiology (AREA)
- Genetics & Genomics (AREA)
- Primary Health Care (AREA)
- Biomedical Technology (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to a cloud manufacturing service combination optimization selection method based on a preference NSGA-III algorithm, which comprises the steps of generating an initial population according to a coding operation method to generate offspring individuals; decoding all individuals in the current population, and calculating a combined service QoS index value according to the decoded individuals; r is selected by adopting a non-dominant sorting method t Dividing into different non-dominant layers, defining F l As critical layer, adopts critical layer individual selection method from F l In selecting K excellent individuals while retaining P t+1 In (2), using a modular factor algorithm, at P t+1 In the process, selectThe individual elite individuals conduct further local searches. According to the invention, decision preference information is introduced into the algorithm in an interactive mode through the self-adaptive construction preference reference points so as to improve the solving efficiency of the algorithm; meanwhile, aiming at the problem of insufficient local searching capability of the algorithm, an evolution mechanism of the algorithm is improved, and a model factor algorithm of mixing genetic searching and local searching is provided so as to improve the solving capability of the algorithm.
Description
Technical Field
The invention relates to the field of cloud manufacturing service scheduling, in particular to a cloud manufacturing service combination optimization selection method based on a preference NSGA-III algorithm.
Background
Cloud manufacturing is a new intelligent manufacturing mode based on cloud computing, internet of things, big data and service-oriented technology, social manufacturing resources are integrated through a network and a cloud platform, so that the resource utilization rate is effectively improved, the production cost is reduced, and personalized service is provided for users. In order to realize centralized management of various manufacturing resources in a heterogeneous environment, a service provider virtualizes and services various manufacturing resources and manufacturing capabilities into manufacturing cloud services and distributes the manufacturing cloud services to a cloud manufacturing service platform, and the cloud manufacturing service platform is responsible for centralized intelligent management and operation of the manufacturing resources. When the platform receives the manufacturing task of the service requester, the task is decomposed into a plurality of subtask units, and all manufacturing services meeting the functional requirements are searched out from the mass cloud services according to the functional requirements of each subtask unit to form a candidate service set. Each candidate service set typically contains a large number of candidate services with different quality of service (Quality of Service, qoS) as an important reference for service requesters to select a combined service, qoS metrics are numerous and conflicting, making service combination optimization selection (Service Composition Optimal Selection, SCOS) a complex NP-Hard Multi-objective optimization problem (Multi-objective Optimization Problems, MOPs) with complex characteristics of large scale, high dimensionality, preference, nonlinearity, combined explosion, etc. The invention mainly considers that under the influence of satisfying the association relation between services, a plurality of QoS indexes are weighted and optimized according to the weight, and finally, an optimal combination service is selected from all combination services to execute tasks.
Currently, various evolutionary algorithms such as genetic algorithm, ant colony algorithm, particle swarm optimization algorithm, teaching-learning optimization algorithm and the like are proposed to solve SCOS problem of specific targets in different scenes. All the above researches adopt a single-target optimization method, all the sub-targets which are optimized are aggregated into a single target by utilizing a normalization method such as a weighting method, so that the multi-target optimization problem is converted into a single-target problem (single objective problems, SOPs), and then the single-target optimization method is adopted for solving. In practical application, because of the mutual conflict of QoS indexes, the service requester is more prone to weighing among a plurality of QoS indexes, the system gives a group of Pareto optimal solution sets meeting the preference of the service requester, and the service requester selects one of the combined services according to own preference. Thus, a multi-objective optimization method is employed herein to simultaneously optimize multiple QoS metrics in a high-dimensional space according to decision preferences, resulting in a set of Pareto optimal solutions that meet the service requester needs. The SCOS problem has the characteristics of preference and high dimensionality, and algorithms which can be used for solving the problem include an r-dominance algorithm, an NSGA-III (Non-dominated Sorting Genetic Algorithm III, NASG-III) and other evolutionary algorithms. The r-dominance algorithm converts the Pareto non-dominant solution set into a strict partial sequence set by adopting a method of setting a reference point, guides the algorithm to search the region (region of interest, ROI) of interest of a decision maker, but has larger dependence on the reference point set by the decision maker, and the performance of the algorithm is seriously affected when the position of the selected reference point is not proper. In the SCOS problem, the optimal front of Pareto is unknown, setting a proper reference point is very difficult, the algorithm needs to calculate the weighted Euclidean distance between the population individuals and the reference point, and when the reference points are more, the algorithm has lower efficiency. The NSGA III algorithm is a high-dimensional multi-objective evolutionary algorithm proposed by Deb and the like, in the environment selection, individuals are selected by adopting a method of constructing reference points, the number of individuals associated with each reference point is emphasized on the basis of a dominant relationship, so that the population has good distribution, but as a global search algorithm, NSGA-III aims at finding the whole Pareto optimal boundary, and SCOS problems only need to obtain a decision maker ROI, so that the algorithm is directly applied to the SCOS problems, the effect is not ideal, and the pressure of the algorithm is small and the convergence is slow because the NASG-III algorithm oversubstanties the distribution of the population. Therefore, in view of the SCOS problem, it is necessary to find an algorithm that can solve the high-dimensional problem and has preference, but there is little research in this respect.
Disclosure of Invention
Aiming at the defects of the prior art, the invention provides a cloud manufacturing service combination optimization selection method based on a preference NSGA-III algorithm, which solves the problem of QoS index balance optimization in service combination optimization selection in a cloud manufacturing environment, thereby improving the solving efficiency and solving capability, helping a service requester to find a group of combination services meeting the preference requirement of the requester, and having engineering guidance significance for scheduling optimization of cloud manufacturing.
The technical scheme adopted by the invention for achieving the purpose is as follows:
a cloud manufacturing service combination optimization selection method based on a preference NSGA-III algorithm comprises the following steps:
step 1: generating an initial population P with a scale of N according to a coding operation method 0 The initial population P 0 Is population P t A state when t=0, where t is an iteration count;
step 2: generating child individuals Q using a global search algorithm in a modular factor algorithm t ；
Step 3: for all individuals R in the current population t Performing a decoding operation, wherein R t ←P t ∪Q t Calculating a combined service QoS index value according to the decoded individual;
step 4: r is selected by adopting a non-dominant sorting method t Divided into different non-dominant layers (F 1 ,F 2 ,…F n ) N is the number of non-dominant layers if |F 1 ∪F 2 ∪…F l-1 |<N and |F 1 ∪F 2 ∪…F l |>N, P t+1 ←F 1 ∪F 2 ∪…F l-1 Definition F l Is a critical layer, 1 of<l<n, executing the step 5; otherwise |F 1 ∪F 2 ∪…F l-1 When |=n, step 6 is performed;
step 5: from F by critical layer individual selection method l In selecting K excellent individuals while retaining P t+1 Wherein K+.N- |P t+1 |；
Step 6: using a modular factor algorithm, at P t+1 In the process, selectFurther local searches are performed on individual elite individuals, where P ls Is a local search probability;
step 7: t++1; ending if t is greater than the maximum iteration count gmax; otherwise, returning to the step 2.
The coding operation method comprises the following steps: the decomposed subtask units are changed into a sequential structure to obtain a subtask unit set Task, wherein each subtask unit corresponds to a candidate resource service set, so that a service resource can be selected from each candidate solution set to form a combined service.
The global search algorithm in the model factor algorithm is as follows:
step 1: in the form of non-replacement, in population P t Two individuals are selected each time to carry out cross operation until the population P is traversed t All individuals in the group; wherein the cross operation is: randomly setting two cross points in two individuals, and exchanging service resources on part of subtask units in the cross points to obtain two cross individuals;
step 2: performing mutation operation on all obtained crossed individuals, namely randomly selecting one or more positions, changing the service resource value of the subtask unit at the position, and generating a child individual Q t Where t is the iteration count.
The decoding operation is as follows:
and integrating and aggregating the subtask units in the coding individuals according to the structures among the subtask units, and calculating according to a combination service formula of the manufacturing cloud service QoS aggregation structure to obtain a combination service QoS index value of the resource.
The QoS index value of the combined service is calculated by selecting QoS index according to the self requirement of the user and establishing an evaluation model.
The critical layer individual selection method comprises the following steps:
the individuals in the population are evenly referenced to the point Z a Performing association operation to obtain that each uniform reference point is in a set P t+1 ＝S t /F l The number of individuals p associated with (a) and then using the construction preference reference point Z r And to critical layer F l The individual retention operation is carried out on the individuals in the population, K excellent individuals are selected, and K is ≡N- |P t+1 |。
The individual retention operation is:
first at the preference reference point set Z r Selecting the reference point j with the least number of associated individuals, wherein the number of associated individuals is ρ j Randomly selecting a j if there are a plurality of individuals;
at F l Searching for an individual I associated with reference point j j If I j I=0, then the reference point is not considered any more in the rest of the operations, it is shifted out of Z r The method comprises the steps of carrying out a first treatment on the surface of the If I j |>1 and ρ j =0, then select the individual nearest to the reference point j to join P t+1 In (a) and (b); if I j |>1 and ρ j 1. Gtoreq., then randomly select one individual to add P t+1 In (a) and (b);
repeating the above operation when the preference points are collected Z r Is empty and |P t+1 |<When N is, Z is selected e As a reference point, where Z e Is Z a And Z is r The above operation is continued until the number of selected individuals is K.
The uniform reference point Z a The method is generated by a boundary crossing construction weight method, namely H reference points which are uniformly generated on a standard hyperplane, wherein H=C (M+P-1, P), M is the number of optimization targets, and P is the number of uniform division parts.
The preference reference point Z r The reference point theta segment is adopted for convergence, namely the weight of each evaluation index is w= { w 1 ，w 2 ，…，w M M is the number of optimization indexes andw i ∈[0,1]r is the total number of experts, gamma r Authority of data entered by expert r, and +.>w ri The specific quantized value of the ith index in the index set given to the r-th expert in the evaluation set corresponds to the QoS index value in the evaluation model, the evaluation set is divided into 5 classes v= { L, ML, M, MH, H }, where L e [0,0.2 ], ML e (0.2, 0.4)],M∈(0.4,0.6],MH∈(0.6,0.8],H∈(0.8,1]The method comprises the steps of carrying out a first treatment on the surface of the The number of segments is P, the maximum evolution algebra is gmax, the current algebra is t, and the preference reference point of the t th generation is Z a The value of each dimension is less than or equal to the union Z of uniform reference points of the values of the corresponding dimension delta r At this time δ= (δ) 1 ,δ 2 ,…,δ M )，δ i The range of values for the reference points in each dimension is defined by +.>Solving for, among othersThe value of P depends on Z r The number of reference points in (a), typically 100<|Z r |<200。
The local search consists of individual selection and neighborhood search for individuals, wherein
The individual selection is carried out by adopting an aggregation function non-return tournament selection mode, so as to obtain a group of elite solutions;
the neighborhood search of the individual is obtained by changing the service resource value of a random subtask unit on the individual code, then the individuals in the population are improved by adopting a 'Best' acceptance criterion, and the current elite solution is updated, so that the combined service with the minimum aggregation function value on the local search path is found as the final improved individual.
The invention has the following beneficial effects and advantages:
1. the method solves the problem of QoS index balance optimization in service combination optimization selection in cloud manufacturing environment.
2. User decision preference information is introduced into the algorithm, so that the solving efficiency of the algorithm is improved, and the satisfaction degree of the service request is improved.
3. And a local search strategy is introduced into the algorithm, so that the solving capability of the algorithm is improved.
Drawings
FIG. 1 is a flow chart of the method of the present invention;
FIG. 2 is a graph of delta values of parameters at different values of θ according to the present invention;
FIG. 3 is a graph of linguistic variables versus a set of reviews in accordance with the present invention;
FIG. 4 is a mapping relationship diagram between a combination service and individual codes;
FIG. 5 is a flow chart of an automobile manufacturing process;
FIG. 6 is a comparison of a box graph of QoS values for a combined service solved by the proposed algorithm and the modified GA algorithm;
fig. 7 is a graph showing the time spent by the algorithm of the present invention and the modified GA algorithm for combining services.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
In order to make the above objects, features and advantages of the present invention more comprehensible, embodiments accompanied with figures are described in detail below. In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention. The invention may be embodied in many other forms than described herein and similarly modified by those skilled in the art without departing from the spirit or scope of the invention, which is therefore not limited to the specific embodiments disclosed below.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention belongs. The terminology used in the description of the invention herein is for the purpose of describing particular embodiments only and is not intended to be limiting of the invention.
A flow chart of the method of the present invention is shown in fig. 1.
The method comprises the following steps:
step 1: generating a primary population P with a scale of N according to a coding operation method t T is the iteration count, where t=0; the method of real number coding is adopted to change the decomposed subtask units into a sequential structure, as shown in fig. 4, a subtask unit set Task is obtained according to the mapping relation between individual codes and combined services, wherein each subtask unit corresponds to a candidate resource service set, therefore, one service resource can be selected in each candidate solution set to form a combined service, and the mth combined service is P m . Then P m The corresponding service resource coding order is the coding order of the individual, so the individual is denoted (m 1 ，m 2 ，m 3 ，…，m i ，…，m n )。
Step 2: generating child individuals Q using a global search algorithm in a modular factor algorithm t ；
The global search algorithm is:
in the form of non-replacement, in population P t Two individuals are selected each time to carry out cross operation until the population P is traversed t All individuals in the group; wherein the cross operation is: randomly setting two cross points in two individuals, and exchanging service resources on part of subtask units in the cross points to obtain two cross individuals;
performing mutation operation on all obtained crossed individuals, namely randomly selecting one or more positions, changing the service resource value of the subtask unit at the position, and generating a child individual Q t Where t is the iteration count.
Step 3: for all individuals R in the current population t ←P t ∪Q t Performing decoding operation, and calculating QoS index values according to the decoded individuals; the individual to be encoded (m 1 ，m 2 ，m 3 ，…，m i ，…，m n ) The subtask units in the cloud service QoS aggregation structure are integrated and aggregated according to the structure between the subtask units, and the combination service QoS value of the resource is calculated according to the combination service formula of the manufacturing cloud service QoS aggregation structure. The method is favorable for fully utilizing the idle time of resources to improve the utilization rate and shorten the task completion time.
The QoS index value includes the following:
service time: t=t delay +T process +T trans Wherein T is delay 、T process And T trans Representing the time required for the service request to be sent to the logistics generated by the accepted time, the service execution time and the spatial geographical position constraint, respectively.
Service cost: c=c service +C trans Wherein C service 、C trans Representing service components including equipment and material costsCost and logistic cost.
Quality of service: p=n success ÷N total Wherein N is success And N total The number of times the service successfully executes the task and the total number of times the service executes the task are respectively represented.
Service robustness: r=n suc ÷N acc Wherein N is suc And N acc The number of times that the service successfully completes the cloud task under the abnormal condition and the total number of times that the service is abnormal are respectively represented.
Service credibility:wherein R is rep (k) Refers to the reputation evaluation value R of the kth service requester for the service rep (k)∈[0,1]Indicating the total number of service evaluations.
Service information feedback capability: i=i process +I trans Wherein I process And I trans And the processing progress feedback time rate and the logistics progress feedback time rate are respectively represented.
Service delivery time rate: d=d success ÷D total Wherein D is success And D total Respectively representing the number of services delivered in time in a certain time and the total required service number in the time.
Service delivery accuracy: a=a success ÷A total Wherein A is success And A total Indicating the number of accurate completed services at a certain time and the total number of service deliveries in this time period, respectively.
Examples:
according to the user's needs, the embodiment selects QoS index values, which are respectively processing time, manufacturing cost, product quality, delay risk and service satisfaction; wherein, the liquid crystal display device comprises a liquid crystal display device,
the processing time consists of two parts, namely the processing time for completing unit tasks by manufacturing resources and the transportation time generated by the circulation of the tasks among different manufacturing resources;
the manufacturing cost is composed of two parts, namely direct production and processing cost and transportation cost;
the product quality is the qualification rate of finishing the processing task;
delay risk, which is the failure rate of manufacturing resources in processing task units;
service satisfaction, which is the evaluation of the satisfaction of the service requester on the provided service.
Step 4: r is selected by adopting a non-dominant sorting method t Divided into different non-dominant layers (F 1 ,F 2 …) and to reserve non-dominant sets of high priority to next generation P t+1 In, when |F 1 ∪F 2 ∪…F l-1 |<N but |F 1 ∪F 2 ∪…F l-1 |>Definition of F when N l Is a critical layer;
step 5: from F using critical layer selection method l In selecting k=n- |p t+1 I excellent individuals were simultaneously retained to P t+1 In (a) and (b); the critical layer selection method is to combine individuals in the population with a uniform reference point Z a Performing association operation to obtain that each uniform reference point is in a set P t+1 ＝S t /F l The number of individuals p associated with (a) and then using the construction preference reference point Z r And to critical layer F l Is subject to individual retention operations. Uniform reference point Z a In order to construct a weight method through boundary crossing, H reference points are uniformly generated on a standard hyperplane, H=C (M+P-1, P), wherein M is the number of optimization targets, and P is the number of uniform partitions. Preference reference point Z r Selecting by means of convergence of reference point theta segment, fig. 2 shows delta values under different theta, and assuming that the weight of each evaluation index is w= { w 1 ，w 2 ，…，w M }，w i ∈[0,1]The number of divided parts is P, the maximum evolution algebra is gmax, and the preference reference point of the t generation is Z a The value of each dimension is less than or equal to the union Z of uniform reference points of the values of the corresponding dimension delta r At this time δ= (δ) 1 ,δ 2 ,…,δ M )，θ is between 2 and 5 according to the weight of the indexThe value between them can be defined according to the user's requirements, or according to the formula + ->Obtaining the product. To balance the search capability and efficiency of the algorithm, the value of P depends on Z r The number of reference points in (a), typically 100<|Z r |<200。
The evaluation weight of each index gives out the specific quantization method of each index through constructing an index set and a judgment set. The set of metrics is a set of target compositions that need to be optimized, e.g., in SCOS problems, the set of metrics can be represented as U= { U 1 ,u 2 ,u 3 ,u 4 ,u 5 }, u therein 1 Time, u 2 Representing the cost, u 3 Representing the quality, u 4 Representing robustness, u 5 Representing reputation. The evaluation set is used for quantifying decision weights and is divided into 5 levels v= { L, ML, M, MH, H }, wherein fig. 3 shows the correspondence between linguistic variables and the evaluation set. And then determining a specific quantitative value of each index in the index set in the judgment set by adopting a Delphi method. Finally, the weights set by expert authority are clustered to obtain the evaluation weight w= { w of the index 1 ,w 2 ,…,w M M is the number of optimization indexes andw i ∈[0,1]r is the total number of experts, gamma r Authority of data entered by expert r, andw ri a specific quantized value in the evaluation set for the ith index is given to the r-th expert.
Individual retention operation: first at the preference reference point set Z r Selecting the reference point j with the least number of associated individuals, wherein the number of associated individuals is ρ j If there are multiple individuals, a j is randomly selected. At F l Searching for an individual I associated with reference point j j If I j I=0, the reference point is not considered any more in the rest of the operations,removing it from Zr; if I j |>1 and ρ j =0, then select the individual nearest to the reference point j to join P t+1 In (a) and (b); if I j |>1 and ρ j 1. Gtoreq., then randomly select one individual to add P t+1 Is a kind of medium. Repeating the above operation when the preference points are collected Z r Is empty but |P t+1 |<When N is, Z is selected a /Z r As a reference point, the above operation is continued until the number of selected individuals is K.
Step 6: using a modular factor algorithm, at P t+1 In the process, selectPerforming further local searches on individual elite individuals; the local search based on the modulo-cause algorithm consists of two parts, individual selection and neighborhood search for an individual. The individual selection is carried out by adopting a mode of selecting an aggregation function without replacing a tournament, a group of elite solutions are obtained, and the local search aggregation function is defined by the following formula:
f(x,λ)＝λ 1 f 1 (x)+λ 2 f 2 (x)+λ 3 f 3 (x)+λ 4 f 4 (x)+λ 5 f 5 (x)
wherein: lambda (lambda) i ＝w i i＝1,2…,5
Wherein w is the evaluation weight of each index, f 1 (x)，f 2 (x)，f 3 (x)，f 4 (x)，f 5 (x) Respectively set as 5 QoS values of individuals after the population self-adaptive standardization. A set of elite solutions is then selected in a championship selection mode without replacement according to the magnitude of the f (x, lambda) value.
The individual neighborhood search adopts the service resource value m of a random subtask unit i on the individual code by changing G i Obtained, i.e. m i ←k，And m is i Not equal to k. Finally, the individuals in the population are refined using the "Best" acceptance criteria, i.e., the current elite solution if and only if f (G', λ) < f (G, λ)G is updated to find the combined service with the smallest f (G, λ) value on the local search path as the final improvement individual.
Step 7: t++1; ending if t is greater than the maximum iteration count gmax; otherwise, returning to the step 2.
And optimizing and selecting application research scenes by taking the manufacturing process of the automobile as a manufacturing cloud service combination of service association perception, and comparing the proposed algorithm with an improved GA algorithm. 10 and 20 subtask units are selected for flow simulation of the real process, respectively. As shown in fig. 5, which is a flow chart of automobile manufacturing, one and two consecutive processes are performed, respectively. All subtasks in the manufacturing process correspond to respective candidate cloud service sets, each candidate cloud service set respectively comprises 20, 50 and 100 candidate cloud services, and the QoS values in the candidate cloud service sets are assumed to be in the value ranges of Q respectively time ∈(0,100]、Q cost ∈(0,20]、Q qua ∈[0.75,0.95]、Q rob ∈[0.75,0.95]、Q rep ∈[0.75,0.95]The decision-preference language variables of the 5 indices are { H, H, ML, ML, ML }, respectively. The parameters in the modified GA algorithm are set as follows: evaluation weight w= (0.35,0.35,0.1,0.1,0.1), population number n=100, maximum genetic algebra gmax=1500, convergence algebra maxgen=1000, crossover probability P c Probability of variation P =0.9 m =0.2. The parameters in the R-NSGA-III algorithm are set as follows: evaluation weight w= (0.9,0.9,0.3,0.3,0.3), population individual number n=150, maximum genetic algebra gmax=600, crossover probability P c Probability of variation P =0.9 m =0.2, local search probability P l ＝0.2。
Fig. 6 shows a box diagram of QoS values of the two kinds of algorithm selected combination services, and fig. 7 shows running time of the two kinds of algorithm selected different combination services, wherein the front of the horizontal line represents the number of subtask units, and the rear of the horizontal line represents the number of candidate cloud services of each subtask unit. Therefore, in the application research problem discussed in this section, the R-NSGA iii algorithm and the improved GA algorithm proposed herein can find an optimal combination service on the premise of meeting the preference information. The R-NSGA III algorithm has good distribution, a group of solutions meeting decision requirements can be obtained for a service requester to select, and the box graph value range comprises solutions solved by the improved GA algorithm, but compared with the improved GA algorithm, the R-NSGA III algorithm has longer time consumption and lower efficiency.
Claims (8)
1. A cloud manufacturing service combination optimization selection method based on a preference NSGA-III algorithm is characterized in that: the method comprises the following steps:
step 1: generating an initial population P with a scale of N according to a coding operation method 0 The initial population P 0 Is population P t A state when t=0, where t is an iteration count;
step 2: generating child individuals Q using a global search algorithm in a modular factor algorithm t ；
Step 3: for all individuals R in the current population t Performing a decoding operation, wherein R t ←P t ∪Q t Calculating a combined service QoS index value according to the decoded individual;
step 4: r is selected by adopting a non-dominant sorting method t Divided into different non-dominant layers (F 1 ,F 2 ,…F n ) N is the number of non-dominant layers if |F 1 ∪F 2 ∪…F l-1 |<N and |F 1 ∪F 2 ∪…F l |>N, P t+1 ←F 1 ∪F 2 ∪…F l-1 Definition F l Is a critical layer, 1 of<l<n, executing the step 5; otherwise |F 1 ∪F 2 ∪…F l-1 When |=n, step 6 is performed;
step 5: from F by critical layer individual selection method l In selecting K excellent individuals while retaining P t+1 Wherein K+.N- |P t+1 |；
The critical layer individual selection method comprises the following steps:
the individuals in the population are evenly referenced to the point Z a Performing association operation to obtain that each uniform reference point is in a set P t+1 ＝S t /F l The number of individuals p associated with (a) and then using the construction preference reference point Z r And to critical layer F l The individual retention operation is carried out on the individual in the list, and K excellent materials are selectedIndividuals, K++N- |P t+1 |；
The individual retention operation is:
first at the preference reference point set Z r Selecting the reference point j with the least number of associated individuals, wherein the number of associated individuals is ρ j Randomly selecting a j if there are a plurality of individuals;
at F l Searching for an individual I associated with reference point j j If I j I=0, then the reference point is not considered any more in the rest of the operations, it is shifted out of Z r The method comprises the steps of carrying out a first treatment on the surface of the If I j |>1 and ρ j =0, then select the individual nearest to the reference point j to join P t+1 In (a) and (b); if I j |>1 and ρ j 1. Gtoreq., then randomly select one individual to add P t+1 In (a) and (b);
repeating the above operation when the preference points are collected Z r Is empty and |P t+1 |<When N is, Z is selected e As a reference point, where Z e Is Z a And Z is r Continuing to repeat the above operation until the number of selected individuals is K;
step 6: using a modular factor algorithm, at P t+1 In the process, selectFurther local searches are performed on individual elite individuals, where P ls Is a local search probability;
step 7: t++1; ending if t is greater than the maximum iteration count gmax; otherwise, returning to the step 2.
2. The cloud manufacturing service composition optimization selection method based on the preference NSGA-iii algorithm according to claim 1, wherein: the coding operation method comprises the following steps: the decomposed subtask units are changed into a sequential structure to obtain a subtask unit set Task, wherein each subtask unit corresponds to a candidate resource service set, so that a service resource can be selected from each candidate solution set to form a combined service.
3. The cloud manufacturing service composition optimization selection method based on the preference NSGA-iii algorithm according to claim 1, wherein: the global search algorithm in the model factor algorithm is as follows:
step 1: in the form of non-replacement, in population P t Two individuals are selected each time to carry out cross operation until the population P is traversed t All individuals in the group; wherein the cross operation is: randomly setting two cross points in two individuals, and exchanging service resources on part of subtask units in the cross points to obtain two cross individuals;
step 2: performing mutation operation on all obtained crossed individuals, namely randomly selecting one or more positions, changing the service resource value of the subtask unit at the position, and generating a child individual Q t Where t is the iteration count.
4. The cloud manufacturing service composition optimization selection method based on the preference NSGA-iii algorithm according to claim 1, wherein: the decoding operation is as follows:
and integrating and aggregating the subtask units in the coding individuals according to the structures among the subtask units, and calculating according to a combination service formula of the manufacturing cloud service QoS aggregation structure to obtain a combination service QoS index value of the resource.
5. The cloud manufacturing service combination optimization selection method based on the preference NSGA-iii algorithm according to claim 1, wherein the combination service QoS index value is calculated by selecting a QoS index according to the user's own needs to build an evaluation model.
6. The cloud manufacturing service composition optimization selection method based on preference NSGA-iii algorithm according to claim 1, wherein the uniform reference point Z a The method is generated by a boundary crossing construction weight method, namely H reference points which are uniformly generated on a standard hyperplane, wherein H=C (M+P-1, P), M is the number of optimization targets, and P is the number of uniform division parts.
7. The preference-based NS of claim 1Cloud manufacturing service combination optimization selection method of GA-III algorithm, wherein preference reference point Z r The reference point theta segment is adopted for convergence, namely the weight of each evaluation index is w= { w 1 ，w 2 ，…，w M M is the number of optimization indexes andw i ∈[0,1]r is the total number of experts, gamma r Authority of data entered by expert r, and +.>w ri The specific quantized value of the ith index in the index set given to the r-th expert in the evaluation set corresponds to the QoS index value in the evaluation model, the evaluation set is divided into 5 classes v= { L, ML, M, MH, H }, where L e [0,0.2 ], ML e (0.2, 0.4)],M∈(0.4,0.6],MH∈(0.6,0.8],H∈(0.8,1]The method comprises the steps of carrying out a first treatment on the surface of the The number of segments is P, the maximum evolution algebra is gmax, the current algebra is t, and the preference reference point of the t th generation is Z a The value of each dimension is less than or equal to the union Z of uniform reference points of the values of the corresponding dimension delta r At this time δ= (δ) 1 ,δ 2 ,…,δ M )，δ i The range of values for the reference points in each dimension is defined by +.>Find out->The value of P depends on Z r The number of reference points in (a), typically 100<|Z r |<200。
8. The cloud manufacturing service composition optimization selection method based on the preference NSGA-iii algorithm according to claim 7, wherein the local search consists of two parts of individual selection and neighborhood search for individual, wherein
The individual selection is carried out by adopting an aggregation function non-return tournament selection mode, so as to obtain a group of elite solutions;
the neighborhood search of the individual is obtained by changing the service resource value of a random subtask unit on the individual code, then the individuals in the population are improved by adopting a 'Best' acceptance criterion, and the current elite solution is updated, so that the combined service with the minimum aggregation function value on the local search path is found as the final improved individual.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811416182.5A CN111222665B (en) | 2018-11-26 | 2018-11-26 | Cloud manufacturing service combination optimization selection method based on preference NSGA-III algorithm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201811416182.5A CN111222665B (en) | 2018-11-26 | 2018-11-26 | Cloud manufacturing service combination optimization selection method based on preference NSGA-III algorithm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN111222665A CN111222665A (en) | 2020-06-02 |
| CN111222665B true CN111222665B (en) | 2023-09-26 |
Family
ID=70828691
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201811416182.5A Active CN111222665B (en) | 2018-11-26 | 2018-11-26 | Cloud manufacturing service combination optimization selection method based on preference NSGA-III algorithm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN111222665B (en) |
Families Citing this family (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN112803984B (en) * | 2020-12-29 | 2023-05-19 | 国网甘肃省电力公司信息通信公司 | Method for switching between satellites in space-earth integrated communication network |
| CN112765407B (en) * | 2020-12-30 | 2022-11-11 | 重庆邮电大学 | QoS service combination method based on user preference in Internet of things environment |
| CN112801387B (en) * | 2021-02-04 | 2022-07-29 | 南京邮电大学 | Cloud manufacturing service combination optimization method based on improved genetic algorithm |
| CN113033100B (en) * | 2021-03-29 | 2022-11-29 | 重庆大学 | Cloud manufacturing service combination method based on hybrid teaching optimization algorithm |
| CN113627716A (en) * | 2021-06-30 | 2021-11-09 | 上海大学 | Anti-interference vehicle scheduling method for multiple distribution centers |
| CN114221990A (en) * | 2021-12-17 | 2022-03-22 | 武汉长江鲲鹏生态创新科技有限公司 | Combined service optimization method under cloud platform |
| CN116307296B (en) * | 2023-05-22 | 2023-09-29 | 南京航空航天大学 | Cloud resource optimization configuration method |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104035816A (en) * | 2014-05-22 | 2014-09-10 | 南京信息工程大学 | Cloud computing task scheduling method based on improved NSGA-II |
| CN105488568A (en) * | 2015-11-30 | 2016-04-13 | 清华大学 | Meme evolution multiobjective optimization scheduling method based on objective importance decomposition |
| CN107346469A (en) * | 2017-06-12 | 2017-11-14 | 哈尔滨理工大学 | Multiple target integrated dispatch method is transported under cloud manufacturing environment more |
| CN107679750A (en) * | 2017-09-30 | 2018-02-09 | 桂林电子科技大学 | A kind of cloud manufacturing service reso urce matching method based on adaptation coefficient genetic algorithm |
| CN108681789A (en) * | 2018-05-09 | 2018-10-19 | 浙江财经大学 | A kind of cloud manufacturing service optimization method |
| CN108805434A (en) * | 2018-05-25 | 2018-11-13 | 河海大学 | A kind of step power station Multiobjective Optimal Operation method based on improvement NSGA- III |
-
2018
- 2018-11-26 CN CN201811416182.5A patent/CN111222665B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104035816A (en) * | 2014-05-22 | 2014-09-10 | 南京信息工程大学 | Cloud computing task scheduling method based on improved NSGA-II |
| CN105488568A (en) * | 2015-11-30 | 2016-04-13 | 清华大学 | Meme evolution multiobjective optimization scheduling method based on objective importance decomposition |
| CN107346469A (en) * | 2017-06-12 | 2017-11-14 | 哈尔滨理工大学 | Multiple target integrated dispatch method is transported under cloud manufacturing environment more |
| CN107679750A (en) * | 2017-09-30 | 2018-02-09 | 桂林电子科技大学 | A kind of cloud manufacturing service reso urce matching method based on adaptation coefficient genetic algorithm |
| CN108681789A (en) * | 2018-05-09 | 2018-10-19 | 浙江财经大学 | A kind of cloud manufacturing service optimization method |
| CN108805434A (en) * | 2018-05-25 | 2018-11-13 | 河海大学 | A kind of step power station Multiobjective Optimal Operation method based on improvement NSGA- III |
Also Published As
| Publication number | Publication date |
|---|---|
| CN111222665A (en) | 2020-06-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN111222665B (en) | Cloud manufacturing service combination optimization selection method based on preference NSGA-III algorithm | |
| Zhu et al. | Order fulfillment cycle time estimation for on-demand food delivery | |
| CN110751293B (en) | Cloud manufacturing multi-task scheduling optimization method based on game theory | |
| CN111124689A (en) | Dynamic allocation method for container resources in cluster | |
| CN112149990B (en) | Fuzzy supply and demand matching method based on prediction | |
| CN104077634B (en) | active-reactive type dynamic project scheduling method based on multi-objective optimization | |
| CN109559033B (en) | Socialized team member optimization method oriented to cloud design and manufacturing mode | |
| CN110321217A (en) | A kind of cloud resource dispatching method, device, equipment and the storage medium of multiple target | |
| CN110098964A (en) | A kind of disposition optimization method based on ant group algorithm | |
| CN109598366B (en) | Optimized scheduling method for takeout delivery process | |
| CN115480876A (en) | Cloud computing task scheduling method and system based on ant colony algorithm optimization | |
| CN112036651A (en) | Electricity price prediction method based on quantum immune optimization BP neural network algorithm | |
| CN111832817A (en) | Small world echo state network time sequence prediction method based on MCP penalty function | |
| CN113139639A (en) | MOMBI-based smart city application-oriented multi-target calculation migration method and device | |
| CN110633784B (en) | Multi-rule artificial bee colony improvement algorithm | |
| CN115271130B (en) | Dynamic scheduling method and system for maintenance order of ship main power equipment | |
| CN113220437B (en) | Workflow multi-target scheduling method and device | |
| Chiang et al. | Multi-mode resource-constrained project scheduling by ant colony optimization with a dynamic tournament strategy | |
| CN116089083A (en) | Multi-target data center resource scheduling method | |
| CN114819660A (en) | Dynamic evolution design crowdsourcing human resource task matching method and system | |
| CN115016889A (en) | Virtual machine optimization scheduling method for cloud computing | |
| CN116339932A (en) | Resource scheduling method, device and server | |
| CN113642808A (en) | Dynamic scheduling method for cloud manufacturing resource change | |
| CN114625497B (en) | Credible service combination method based on cooperative sensing | |
| Zhang et al. | Multi-Criteria Client Selection and Scheduling with Fairness Guarantee for Federated Learning Service |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| CB02 | Change of applicant information |
Address after: Nanping Road in Dongling District of Shenyang city of Liaoning Province, No. 110168 16-2 Applicant after: Shenyang Zhongke CNC Technology Co.,Ltd. Address before: Nanping Road in Dongling District of Shenyang city of Liaoning Province, No. 110168 16-2 Applicant before: Shenyang Golding Nc Intelligence Tech.co.,ltd. |
|
| CB02 | Change of applicant information | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |